Teaching Dialogue In Narrative Writing

Why Your Dialogue Sounds Like It Was Written By A Robot

I spent three years as a developmental editor before moving into teaching. The single most consistent problem I see is that writers treat dialogue as information transfer rather than character behavior. Students hand me pages where characters explain their motivations to each other in perfectly grammatical sentences, and nobody talks like that. People interrupt. People dodge. People say the wrong thing on purpose. The core technique I use starts with removing exposition from conversation entirely. I give students a scene where two characters need to discuss something important, then I have them rewrite it with three constraints: no character can state their true goal outright, at least one person must avoid the topic when they have the chance, and every line has to do double duty by revealing relationship dynamics. This usually forces genuine subtext into the writing within a single revision cycle. The scenes go from sterile to usable in about 20 minutes instead of getting stuck in second-draft purgatory where they look fine on paper but feel hollow when read aloud. Here is what that looks like in practice. A standard student draft reads like this:

"I'm afraid you're going to leave me," Sarah said. "I haven't said anything about leaving," Mark replied. "But you will. You've been distant for weeks. I can tell you're planning to go."

After applying the constraints, it becomes something closer to this: Mark set his coffee cup down too hard. The kitchen floor was still wet from this morning's spill and he kept looking at it. "You want the good silver or just the regular?" Sarah asked.

"Whatever's easier," he said, not looking up. "Right. Whatever's easiest." Both characters are talking about exactly what they mean without either of them saying it. Sarah is registering his withdrawal. Mark is already mentally elsewhere. The wet floor matters because he can't face looking at her directly yet.

One edge case that trips up almost everyone: when two characters share the same emotional baseline. If both are equally angry, equally sad, or equally excited, the conversation flattens into parallel monologues. They reinforce each other instead of creating friction. I had a student once write a entire breakup scene where both characters were devastatingly articulate about their pain. It read like a therapy session transcript, not a moment of rupture. The fix was giving one character the impulse to deflect with humor while the other couldn't stop digging. The clash between those instincts created the scene's energy. Without that asymmetry, the dialogue had nowhere to go.

Subtext Is Not Hiding What Characters Feel

Beginners think subtext means characters are concealing their emotions. It's more useful to think of subtext as characters having competing priorities in the same conversation. One person wants resolution. Another wants escape. A third wants to win. When everyone at the table is playing a different game, the dialogue writes itself because each line becomes a move rather than a statement of feeling. This connects to something called transactional analysis, which is just a fancy way of describing what power shifts happen moment to moment between speakers. In a healthy dialogue sequence, someone should gain ground, lose ground, or shift the terms of the argument at least twice per page. If the power balance stays flat for three pages, readers stop tracking what is happening between the characters and start noticing that the prose is carrying weight it shouldn't need. I also push my students to read dialogue aloud before they consider it finished. Not perform it. Just read it at normal conversation speed. You will catch things immediately that your eyes skip over on the screen. Run-on sentences that should be broken up. Lines where a character would physically need to breathe. Places where the rhythm sounds like an essay rather than speech. This step takes about 90 seconds per page and catches roughly half the problems that normally show up in beta reader feedback.

There is a limit to how much this approach works though. In genre fiction where plot machinery needs to move fast, sometimes characters have to deliver information efficiently. Long, layered subtextual scenes don't belong in every chapter. If you're writing a thriller and the protagonist just found a bomb under the car, having them dance around the revelation for three pages would be a mistake. Know when directness serves the story instead of treating it as a failure of craft. The students who learn that distinction early tend to produce work that feels intentional rather than inconsistent. Another common trap is over-tagging dialogue beats. Every line followed by a gesture or action slows reading velocity to a crawl. It also signals that the writer doesn't trust the dialogue to carry meaning on its own. Use tags sparingly. "He said" and "she said" are invisible to most readers and that invisibility is a feature, not a bug. The eye slides over them and stays on the content. Replace them with action beats only when the beat adds information that the line itself does not convey. If you want a practical exercise, take a published scene you admire and copy it by hand. Not type it. Write it out in longhand. You will feel the rhythm differently when your hand moves at writing speed instead of reading speed. Notice where the author lets silence do the work. Notice where they break a natural sentence in half to create emphasis. This takes about 15 minutes and gives you a physical sense of pacing that no amount of explanation can replicate.